Emergence of fractal geometries in the evolution of a metabolic enzyme.

Fractals are patterns that are self-similar across multiple length-scales1. Macroscopic fractals are common in nature2-4; however, so far, molecular assembly into fractals is restricted to synthetic systems5-12. Here we report the discovery of a natural protein, citrate synthase from the cyanobacterium Synechococcus elongatus, which self-assembles into Sierpinski triangles. Using cryo-electron microscopy, we reveal how the fractal assembles from a hexameric building block. Although different stimuli modulate the formation of fractal complexes and these complexes can regulate the enzymatic activity of citrate synthase in vitro, the fractal may not serve a physiological function in vivo. We use ancestral sequence reconstruction to retrace how the citrate synthase fractal evolved from non-fractal precursors, and the results suggest it may have emerged as a harmless evolutionary accident. Our findings expand the space of possible protein complexes and demonstrate that intricate and regulatable assemblies can evolve in a single substitution.

A hydrophobic ratchet entrenches molecular complexes.

Most proteins assemble into multisubunit complexes1. The persistence of these complexes across evolutionary time is usually explained as the result of natural selection for functional properties that depend on multimerization, such as intersubunit allostery or the capacity to do mechanical work2. In many complexes, however, multimerization does not enable any known function3. An alternative explanation is that multimers could become entrenched if substitutions accumulate that are neutral in multimers but deleterious in monomers; purifying selection would then prevent reversion to the unassembled form, even if assembly per se does not enhance biological function3-7. Here we show that a hydrophobic mutational ratchet systematically entrenches molecular complexes. By applying ancestral protein reconstruction and biochemical assays to the evolution of steroid hormone receptors, we show that an ancient hydrophobic interface, conserved for hundreds of millions of years, is entrenched because exposure of this interface to solvent reduces protein stability and causes aggregation, even though the interface makes no detectable contribution to function. Using structural bioinformatics, we show that a universal mutational propensity drives sites that are buried in multimeric interfaces to accumulate hydrophobic substitutions to levels that are not tolerated in monomers. In a database of hundreds of families of multimers, most show signatures of long-term hydrophobic entrenchment. It is therefore likely that many protein complexes persist because a simple ratchet-like mechanism entrenches them across evolutionary time, even when they are functionally gratuitous.

High-Performance Molecular Dynamics Simulations for Native Mass Spectrometry of Large Protein Complexes with the Fast Multipole Method.

Native mass spectrometry (MS) is widely employed to study the structures and assemblies of proteins ranging from small monomers to megadalton complexes. Molecular dynamics (MD) simulation is a useful complement as it provides the spatial detail that native MS cannot offer. However, MD simulations performed in the gas phase have suffered from rapidly increasing computational costs with the system size. The primary bottleneck is the calculation of electrostatic forces, which are effective over long distances and must be explicitly computed for each atom pair, precluding efficient use of methods traditionally used to accelerate condensed-phase simulations. As a result, MD simulations have been unable to match the capacity of MS in probing large multimeric protein complexes. Here, we apply the fast multipole method (FMM) for computing the electrostatic forces, recently implemented by Kohnke et al. (J. Chem. Theory Comput., 2020, 16, 6938-6949), showing that it significantly enhances the performance of gas-phase simulations of large proteins. We assess how to achieve adequate accuracy and optimal performance with FMM, finding that it expands the accessible size range and time scales dramatically. Additionally, we simulate a 460 kDa ferritin complex over microsecond time scales, alongside complementary ion mobility (IM)-MS experiments, uncovering conformational changes that are not apparent from the IM-MS data alone.

Collision induced unfolding and molecular dynamics simulations of norovirus capsid dimers reveal strain-specific stability profiles.

Collision induced unfolding (CIU) is a method used with ion mobility mass spectrometry to examine protein structures and their stability. Such experiments yield information about higher order protein structures, yet are unable to provide details about the underlying processes. That information can however be provided using molecular dynamics simulations. Here, we investigate the gas-phase unfolding of norovirus capsid dimers from the Norwalk and Kawasaki strains by employing molecular dynamics simulations over a range of temperatures, representing different levels of activation, together with CIU experiments. The dimers have highly similar structures, but their CIU reveals different stability that can be explained by the different dynamics that arises in response to the activation seen in the simulations, including a part of the sequence with previously observed strain-specific dynamics in solution. Our findings show how similar protein variants can be examined using mass spectrometric techniques in conjunction with atomistic molecular dynamics simulations to reveal differences in stability as well as differences in how and where unfolding takes place upon activation.

Enhanced EMC-Advantages of partially known orientations in x-ray single particle imaging.

Single particle imaging of proteins in the gas phase with x-ray free-electron lasers holds great potential to study fast protein dynamics, but is currently limited by weak and noisy data. A further challenge is to discover the proteins' orientation as each protein is randomly oriented when exposed to x-rays. Algorithms such as the expand, maximize, and compress (EMC) exist that can solve the orientation problem and reconstruct the three-dimensional diffraction intensity space, given sufficient measurements. If information about orientation were known, for example, by using an electric field to orient the particles, the reconstruction would benefit and potentially reach better results. We used simulated diffraction experiments to test how the reconstructions from EMC improve with particles' orientation to a preferred axis. Our reconstructions converged to correct maps of the three-dimensional diffraction space with fewer measurements if biased orientation information was considered. Even for a moderate bias, there was still significant improvement. Biased orientations also substantially improved the results in the case of missing central information, in particular in the case of small datasets. The effects were even more significant when adding a background with 50% the strength of the averaged diffraction signal photons to the diffraction patterns, sometimes reducing the data requirement for convergence by a factor of 10. This demonstrates the usefulness of having biased orientation information in single particle imaging experiments, even for a weaker bias than what was previously known. This could be a key component in overcoming the problems with background noise that currently plague these experiments.

Mass Spectrometry and Machine Learning Reveal Determinants of Client Recognition by Antiamyloid Chaperones.

The assembly of proteins and peptides into amyloid fibrils is causally linked to serious disorders such as Alzheimer's disease. Multiple proteins have been shown to prevent amyloid formation in vitro and in vivo, ranging from highly specific chaperone-client pairs to completely nonspecific binding of aggregation-prone peptides. The underlying interactions remain elusive. Here, we turn to the machine learning-based structure prediction algorithm AlphaFold2 to obtain models for the nonspecific interactions of ß-lactoglobulin, transthyretin, or thioredoxin 80 with the model amyloid peptide amyloid ß and the highly specific complex between the BRICHOS chaperone domain of C-terminal region of lung surfactant protein C and its polyvaline target. Using a combination of native mass spectrometry (MS) and ion mobility MS, we show that nonspecific chaperoning is driven predominantly by hydrophobic interactions of amyloid ß with hydrophobic surfaces in ß-lactoglobulin, transthyretin, and thioredoxin 80, and in part regulated by oligomer stability. For C-terminal region of lung surfactant protein C, native MS and hydrogen-deuterium exchange MS reveal that a disordered region recognizes the polyvaline target by forming a complementary ß-strand. Hence, we show that AlphaFold2 and MS can yield atomistic models of hard-to-capture protein interactions that reveal different chaperoning mechanisms based on separate ligand properties and may provide possible clues for specific therapeutic intervention.

Liquid-Liquid Phase Separation Primes Spider Silk Proteins for Fiber Formation via a Conditional Sticker Domain.

Many protein condensates can convert to fibrillar aggregates, but the underlying mechanisms are unclear. Liquid-liquid phase separation (LLPS) of spider silk proteins, spidroins, suggests a regulatory switch between both states. Here, we combine microscopy and native mass spectrometry to investigate the influence of protein sequence, ions, and regulatory domains on spidroin LLPS. We find that salting out-effects drive LLPS via low-affinity stickers in the repeat domains. Interestingly, conditions that enable LLPS simultaneously cause dissociation of the dimeric C-terminal domain (CTD), priming it for aggregation. Since the CTD enhances LLPS of spidroins but is also required for their conversion into amyloid-like fibers, we expand the stickers and spacers-model of phase separation with the concept of folded domains as conditional stickers that represent regulatory units.

Mass Spectrometry of RNA-Binding Proteins during Liquid-Liquid Phase Separation Reveals Distinct Assembly Mechanisms and Droplet Architectures.

Liquid-liquid phase separation (LLPS) of heterogeneous ribonucleoproteins (hnRNPs) drives the formation of membraneless organelles, but structural information about their assembled states is still lacking. Here, we address this challenge through a combination of protein engineering, native ion mobility mass spectrometry, and molecular dynamics simulations. We used an LLPS-compatible spider silk domain and pH changes to control the self-assembly of the hnRNPs FUS, TDP-43, and hCPEB3, which are implicated in neurodegeneration, cancer, and memory storage. By releasing the proteins inside the mass spectrometer from their native assemblies, we could monitor conformational changes associated with liquid-liquid phase separation. We find that FUS monomers undergo an unfolded-to-globular transition, whereas TDP-43 oligomerizes into partially disordered dimers and trimers. hCPEB3, on the other hand, remains fully disordered with a preference for fibrillar aggregation over LLPS. The divergent assembly mechanisms revealed by ion mobility mass spectrometry of soluble protein species that exist under LLPS conditions suggest structurally distinct complexes inside liquid droplets that may impact RNA processing and translation depending on biological context.

Coherent diffractive imaging of proteins and viral capsids: simulating MS SPIDOC.

MS SPIDOC is a novel sample delivery system designed for single (isolated) particle imaging at X-ray Free-Electron Lasers that is adaptable towards most large-scale facility beamlines. Biological samples can range from small proteins to MDa particles. Following nano-electrospray ionization, ionic samples can be m/z-filtered and structurally separated before being oriented at the interaction zone. Here, we present the simulation package developed alongside this prototype. The first part describes how the front-to-end ion trajectory simulations have been conducted. Highlighted is a quadrant lens; a simple but efficient device that steers the ion beam within the vicinity of the strong DC orientation field in the interaction zone to ensure spatial overlap with the X-rays. The second part focuses on protein orientation and discusses its potential with respect to diffractive imaging methods. Last, coherent diffractive imaging of prototypical T = 1 and T = 3 norovirus capsids is shown. We use realistic experimental parameters from the SPB/SFX instrument at the European XFEL to demonstrate that low-resolution diffractive imaging data (q < 0.3 nm-1) can be collected with only a few X-ray pulses. Such low-resolution data are sufficient to distinguish between both symmetries of the capsids, allowing to probe low abundant species in a beam if MS SPIDOC is used as sample delivery.

Structural Basis for Dityrosine-Mediated Inhibition of α-Synuclein Fibrillization.

α-Synuclein (α-Syn) is an intrinsically disordered protein which self-assembles into highly organized ß-sheet structures that accumulate in plaques in brains of Parkinson's disease patients. Oxidative stress influences α-Syn structure and self-assembly; however, the basis for this remains unclear. Here we characterize the chemical and physical effects of mild oxidation on monomeric α-Syn and its aggregation. Using a combination of biophysical methods, small-angle X-ray scattering, and native ion mobility mass spectrometry, we find that oxidation leads to formation of intramolecular dityrosine cross-linkages and a compaction of the α-Syn monomer by a factor of √2. Oxidation-induced compaction is shown to inhibit ordered self-assembly and amyloid formation by steric hindrance, suggesting an important role of mild oxidation in preventing amyloid formation.

Protein orientation in time-dependent electric fields: orientation before destruction.

Proteins often have nonzero electric dipole moments, making them interact with external electric fields and offering a means for controlling their orientation. One application that is known to benefit from orientation control is single-particle imaging with x-ray free-electron lasers, in which diffraction is recorded from proteins in the gas phase to determine their structures. To this point, theoretical investigations into this phenomenon have assumed that the field experienced by the proteins is constant or a perfect step function, whereas any real-world pulse will be smooth. Here, we explore the possibility of orienting gas-phase proteins using time-dependent electric fields. We performed ab initio simulations to estimate the field strength required to break protein bonds, with 45 V/nm as a breaking point value. We then simulated ubiquitin in time-dependent electric fields using classical molecular dynamics. The minimal field strength required for orientation within 10 ns was on the order of 0.5 V/nm. Although high fields can be destructive for the structure, the structures in our simulations were preserved until orientation was achieved regardless of field strength, a principle we denote "orientation before destruction."

Glycan-Induced Protein Dynamics in Human Norovirus P Dimers Depend on Virus Strain and Deamidation Status.

Noroviruses are the major cause of viral gastroenteritis and re-emerge worldwide every year, with GII.4 currently being the most frequent human genotype. The norovirus capsid protein VP1 is essential for host immune response. The P domain mediates cell attachment via histo blood-group antigens (HBGAs) in a strain-dependent manner but how these glycan-interactions actually relate to cell entry remains unclear. Here, hydrogen/deuterium exchange mass spectrometry (HDX-MS) is used to investigate glycan-induced protein dynamics in P dimers of different strains, which exhibit high structural similarity but different prevalence in humans. While the almost identical strains GII.4 Saga and GII.4 MI001 share glycan-induced dynamics, the dynamics differ in the emerging GII.17 Kawasaki 308 and rare GII.10 Vietnam 026 strain. The structural aspects of glycan binding to fully deamidated GII.4 P dimers have been investigated before. However, considering the high specificity and half-life of N373D under physiological conditions, large fractions of partially deamidated virions with potentially altered dynamics in their P domains are likely to occur. Therefore, we also examined glycan binding to partially deamidated GII.4 Saga and GII.4 MI001 P dimers. Such mixed species exhibit increased exposure to solvent in the P dimer upon glycan binding as opposed to pure wildtype. Furthermore, deamidated P dimers display increased flexibility and a monomeric subpopulation. Our results indicate that glycan binding induces strain-dependent structural dynamics, which are further altered by N373 deamidation, and hence hint at a complex role of deamidation in modulating glycan-mediated cell attachment in GII.4 strains.

Predicting the Shapes of Protein Complexes through Collision Cross Section Measurements and Database Searches.

In structural biology, collision cross sections (CCSs) from ion mobility mass spectrometry (IM-MS) measurements are routinely compared to computationally or experimentally derived protein structures. Here, we investigate whether CCS data can inform about the shape of a protein in the absence of specific reference structures. Analysis of the proteins in the CCS database shows that protein complexes with low apparent densities are structurally more diverse than those with a high apparent density. Although assigning protein shapes purely on CCS data is not possible, we find that we can distinguish oblate- and prolate-shaped protein complexes by using the CCS, molecular weight, and oligomeric states to mine the Protein Data Bank (PDB) for potentially similar protein structures. Furthermore, comparing the CCS of a ferritin cage to the solution structures in the PDB reveals significant deviations caused by structural collapse in the gas phase. We then apply the strategy to an integral membrane protein by comparing the shapes of a prokaryotic and a eukaryotic sodium/proton antiporter homologue. We conclude that mining the PDB with IM-MS data is a time-effective way to derive low-resolution structural models.

Software Requirements for the Analysis and Interpretation of Native Ion Mobility Mass Spectrometry Data.

The past few years have seen a dramatic increase in applications of native mass and ion mobility spectrometry, especially for the study of proteins and protein complexes. This increase has been catalyzed by the availability of commercial instrumentation capable of carrying out such analyses. As in most fields, however, the software to process the data generated from new instrumentation lags behind. Recently, a number of research groups have started addressing this by developing software, but further improvements are still required in order to realize the full potential of the data sets generated. In this perspective, we describe practical aspects as well as challenges in processing native mass spectrometry (MS) and ion mobility-MS data sets and provide a brief overview of currently available tools. We then set out our vision of future developments that would bring the community together and lead to the development of a common platform to expedite future computational developments, provide standardized processing approaches, and serve as a location for the deposition of data for this emerging field. This perspective has been written by members of the European Cooperation in Science and Technology Action on Native MS and Related Methods for Structural Biology (EU COST Action BM1403) as an introduction to the software tools available in this area. It is intended to serve as an overview for newcomers and to stimulate discussions in the community on further developments in this field, rather than being an in-depth review. Our complementary perspective (http://dx.doi.org/10.1021/acs.analchem.9b05791) focuses on computational approaches used in this field.

Computational Strategies and Challenges for Using Native Ion Mobility Mass Spectrometry in Biophysics and Structural Biology.

Native mass spectrometry (MS) allows the interrogation of structural aspects of macromolecules in the gas phase, under the premise of having initially maintained their solution-phase noncovalent interactions intact. In the more than 25 years since the first reports, the utility of native MS has become well established in the structural biology community. The experimental and technological advances during this time have been rapid, resulting in dramatic increases in sensitivity, mass range, resolution, and complexity of possible experiments. As experimental methods have improved, there have been accompanying developments in computational approaches for analyzing and exploiting the profusion of MS data in a structural and biophysical context. In this perspective, we consider the computational strategies currently being employed by the community, aspects of best practice, and the challenges that remain to be addressed. Our perspective is based on discussions within the European Cooperation in Science and Technology Action on Native Mass Spectrometry and Related Methods for Structural Biology (EU COST Action BM1403), which involved participants from across Europe and North America. It is intended not as an in-depth review but instead to provide an accessible introduction to and overview of the topic-to inform newcomers to the field and stimulate discussions in the community about addressing existing challenges. Our complementary perspective (http://dx.doi.org/10.1021/acs.analchem.9b05792) focuses on software tools available to help researchers tackle some of the challenges enumerated here.

MkVsites: A tool for creating GROMACS virtual sites parameters to increase performance in all-atom molecular dynamics simulations.

The absolute performance of any all-atom molecular dynamics simulation is typically limited by the length of the individual timesteps taken when integrating the equations of motion. In the GROMACS simulation software, it has for a long time been possible to use so-called virtual sites to increase the length of the timestep, resulting in a large gain of simulation efficiency. Up until now, support for this approach has in practice been limited to the standard 20 amino acids however, shrinking the applicability domain of virtual sites. MkVsites is a set of python tools which provides a convenient way to obtain all parameters necessary to use virtual sites for virtually any molecules in a simulation. Required as input to MkVsites is the molecular topology of the molecule(s) in question, along with a specification of where to find the parent force field. As such, MkVsites can be a very valuable tool suite for anyone who is routinely using GROMACS for the simulation of molecular systems.

Effects of Detergent Micelles on Lipid Binding to Proteins in Electrospray Ionization Mass Spectrometry.

A wide variety of biological processes rely upon interactions between proteins and lipids, ranging from molecular transport to the organization of the cell membrane. It was recently established that electrospray ionization mass spectrometry (ESI-MS) is capable of capturing transient interactions between membrane proteins and their lipid environment, and a detailed understanding of the underlying processes is therefore of high importance. Here, we apply ESI-MS to investigate the factors that govern complex formation in solution and gas phases by comparing nonselective lipid binding with soluble and membrane proteins. We find that exogenously added lipids did not bind to soluble proteins, suggesting that lipids have a low propensity to form electrospray ionization adducts. The presence of detergents at increasing micelle concentrations, on the other hand, resulted in moderate lipid binding to soluble proteins. A direct ESI-MS comparison of lipid binding to the soluble protein serum albumin and to the integral membrane protein NapA shows that soluble proteins acquire fewer lipid adducts. Our results suggest that protein-lipid complexes form via contacts between proteins and mixed lipid/detergent micelles. For soluble proteins, these complexes arise from nonspecific contacts between the protein and detergent/lipid micelles in the electrospray droplet. For membrane proteins, lipids are incorporated into the surrounding micelle in solution, and complex formation occurs independently of the ESI process. We conclude that the lipids in the resulting complexes interact predominantly with sites located in the transmembrane segments, resulting in nativelike complexes that can be interrogated by MS.

Transcription-factor binding and sliding on DNA studied using micro- and macroscopic models.

Transcription factors search for specific operator sequences by alternating rounds of 3D diffusion with rounds of 1D diffusion (sliding) along the DNA. The details of such sliding have largely been beyond direct experimental observation. For this purpose we devised an analytical formulation of umbrella sampling along a helical coordinate, and from extensive and fully atomistic simulations we quantified the free-energy landscapes that underlie the sliding dynamics and dissociation kinetics for the LacI dimer. The resulting potential of mean force distributions show a fine structure with an amplitude of 1 k(B)T for sliding and 12 k(B)T for dissociation. Based on the free-energy calculations the repressor slides in close contact with DNA for 8 bp on average before making a microscopic dissociation. By combining the microscopic molecular-dynamics calculations with Brownian simulation including rotational diffusion from the microscopically dissociated state we estimate a macroscopic residence time of 48 ms at the same DNA segment and an in vitro sliding distance of 240 bp. The sliding distance is in agreement with previous in vitro sliding-length estimates. The in vitro prediction for the macroscopic residence time also compares favorably to what we measure by single-molecule imaging of nonspecifically bound fluorescently labeled LacI in living cells. The investigation adds to our understanding of transcription-factor search kinetics and connects the macro-/mesoscopic rate constants to the microscopic dynamics.

Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles.

Interpretation of mass spectra is challenging because they report a ratio of two physical quantities, mass and charge, which may each have multiple components that overlap in m/z. Previous approaches to disentangling the two have focused on peak assignment or fitting. However, the former struggle with complex spectra, and the latter are generally computationally intensive and may require substantial manual intervention. We propose a new data analysis approach that employs a Bayesian framework to separate the mass and charge dimensions. On the basis of this approach, we developed UniDec (Universal Deconvolution), software that provides a rapid, robust, and flexible deconvolution of mass spectra and ion mobility-mass spectra with minimal user intervention. Incorporation of the charge-state distribution in the Bayesian prior probabilities provides separation of the m/z spectrum into its physical mass and charge components. We have evaluated our approach using systems of increasing complexity, enabling us to deduce lipid binding to membrane proteins, to probe the dynamics of subunit exchange reactions, and to characterize polydispersity in both protein assemblies and lipoprotein Nanodiscs. The general utility of our approach will greatly facilitate analysis of ion mobility and mass spectra.

Charging of DNA Complexes in Positive-Mode Native Electrospray Ionization Mass Spectrometry.

Native mass spectrometry (nMS) provides insights into the structures and dynamics of biomacromolecules in their native-like states by preserving noncovalent interactions through "soft" electrospray ionization (ESI). For native proteins, the number of charges that are acquired scales with the surface area and mass. Here, we explore the effect of highly negatively charged DNA on the ESI charge of protein complexes and find a reduction of the mass-to-charge ratio as well as a greater variation. The charge state distributions of pure DNA assemblies show a lower mass-to-charge ratio than proteins due to their greater density in the gas phase, whereas the charge of protein-DNA complexes can additionally be influenced by the distribution of the ESI charges, ion pairing events, and collapse of the DNA components. Our findings suggest that structural features of protein-DNA complexes can result in lower charge states than expected for proteins.